Training in progress, step 2500, checkpoint
Browse files- checkpoint-2500/config.json +40 -0
- checkpoint-2500/generation_config.json +12 -0
- checkpoint-2500/model.safetensors +3 -0
- checkpoint-2500/optimizer.pt +3 -0
- checkpoint-2500/rng_state.pth +3 -0
- checkpoint-2500/scheduler.pt +3 -0
- checkpoint-2500/trainer_state.json +1783 -0
- checkpoint-2500/training_args.bin +3 -0
checkpoint-2500/config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "meta-llama/Llama-3.2-1B-Instruct",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 128000,
|
| 9 |
+
"eos_token_id": [
|
| 10 |
+
128001,
|
| 11 |
+
128008,
|
| 12 |
+
128009
|
| 13 |
+
],
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 8192,
|
| 19 |
+
"max_position_embeddings": 131072,
|
| 20 |
+
"mlp_bias": false,
|
| 21 |
+
"model_type": "llama",
|
| 22 |
+
"num_attention_heads": 32,
|
| 23 |
+
"num_hidden_layers": 16,
|
| 24 |
+
"num_key_value_heads": 8,
|
| 25 |
+
"pretraining_tp": 1,
|
| 26 |
+
"rms_norm_eps": 1e-05,
|
| 27 |
+
"rope_scaling": {
|
| 28 |
+
"factor": 32.0,
|
| 29 |
+
"high_freq_factor": 4.0,
|
| 30 |
+
"low_freq_factor": 1.0,
|
| 31 |
+
"original_max_position_embeddings": 8192,
|
| 32 |
+
"rope_type": "llama3"
|
| 33 |
+
},
|
| 34 |
+
"rope_theta": 500000.0,
|
| 35 |
+
"tie_word_embeddings": true,
|
| 36 |
+
"torch_dtype": "bfloat16",
|
| 37 |
+
"transformers_version": "4.46.3",
|
| 38 |
+
"use_cache": true,
|
| 39 |
+
"vocab_size": 128256
|
| 40 |
+
}
|
checkpoint-2500/generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128008,
|
| 7 |
+
128009
|
| 8 |
+
],
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.46.3"
|
| 12 |
+
}
|
checkpoint-2500/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f2f34ff678a8eba926b0be9e0ae8380aefdf705c91393adac7f10f3e84ddf04
|
| 3 |
+
size 2471645608
|
checkpoint-2500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5494db8cc9ad65f4a2f678f9aa10c185f5267563360df17f2d6a596fcfc5581b
|
| 3 |
+
size 3724602
|
checkpoint-2500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:775577108bfc56915ff45cab95bf1f0dc44959b9aa4faeaf34dd0014020ace7a
|
| 3 |
+
size 14244
|
checkpoint-2500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3d944f5855dedd93de5810fb83351c896ba8bcf8ae6abf8ac51fb2013da8b6d
|
| 3 |
+
size 1064
|
checkpoint-2500/trainer_state.json
ADDED
|
@@ -0,0 +1,1783 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.813542688910697,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.03925417075564279,
|
| 13 |
+
"grad_norm": 81.5,
|
| 14 |
+
"learning_rate": 8.000000000000001e-07,
|
| 15 |
+
"loss": 0.3818,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.07850834151128558,
|
| 20 |
+
"grad_norm": 322.0,
|
| 21 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 22 |
+
"loss": 0.367,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.11776251226692837,
|
| 27 |
+
"grad_norm": 784.0,
|
| 28 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 29 |
+
"loss": 0.4472,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.15701668302257116,
|
| 34 |
+
"grad_norm": 350.0,
|
| 35 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 36 |
+
"loss": 0.6806,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.19627085377821393,
|
| 41 |
+
"grad_norm": 186.0,
|
| 42 |
+
"learning_rate": 4.000000000000001e-06,
|
| 43 |
+
"loss": 0.9139,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.23552502453385674,
|
| 48 |
+
"grad_norm": 199.0,
|
| 49 |
+
"learning_rate": 4.800000000000001e-06,
|
| 50 |
+
"loss": 0.8205,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.2747791952894995,
|
| 55 |
+
"grad_norm": 118.0,
|
| 56 |
+
"learning_rate": 5.600000000000001e-06,
|
| 57 |
+
"loss": 0.9017,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.3140333660451423,
|
| 62 |
+
"grad_norm": 616.0,
|
| 63 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 64 |
+
"loss": 3.1966,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.35328753680078506,
|
| 69 |
+
"grad_norm": 692.0,
|
| 70 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 71 |
+
"loss": 1.0627,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.39254170755642787,
|
| 76 |
+
"grad_norm": 75.5,
|
| 77 |
+
"learning_rate": 8.000000000000001e-06,
|
| 78 |
+
"loss": 7.3782,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.43179587831207067,
|
| 83 |
+
"grad_norm": 908.0,
|
| 84 |
+
"learning_rate": 8.8e-06,
|
| 85 |
+
"loss": 1.1091,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.47105004906771347,
|
| 90 |
+
"grad_norm": 61.5,
|
| 91 |
+
"learning_rate": 9.600000000000001e-06,
|
| 92 |
+
"loss": 0.9262,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.5103042198233563,
|
| 97 |
+
"grad_norm": 45.25,
|
| 98 |
+
"learning_rate": 1.04e-05,
|
| 99 |
+
"loss": 0.6774,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.549558390578999,
|
| 104 |
+
"grad_norm": 39.75,
|
| 105 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 106 |
+
"loss": 0.8204,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.5888125613346418,
|
| 111 |
+
"grad_norm": 38.0,
|
| 112 |
+
"learning_rate": 1.2e-05,
|
| 113 |
+
"loss": 0.6443,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.6280667320902846,
|
| 118 |
+
"grad_norm": 31.25,
|
| 119 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 120 |
+
"loss": 0.607,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.6673209028459274,
|
| 125 |
+
"grad_norm": 33.5,
|
| 126 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 127 |
+
"loss": 0.674,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.7065750736015701,
|
| 132 |
+
"grad_norm": 37.5,
|
| 133 |
+
"learning_rate": 1.4400000000000001e-05,
|
| 134 |
+
"loss": 0.6806,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.745829244357213,
|
| 139 |
+
"grad_norm": 36.25,
|
| 140 |
+
"learning_rate": 1.5200000000000002e-05,
|
| 141 |
+
"loss": 0.6131,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.7850834151128557,
|
| 146 |
+
"grad_norm": 43.25,
|
| 147 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 148 |
+
"loss": 0.5914,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.8243375858684985,
|
| 153 |
+
"grad_norm": 27.75,
|
| 154 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 155 |
+
"loss": 0.5488,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.8635917566241413,
|
| 160 |
+
"grad_norm": 47.75,
|
| 161 |
+
"learning_rate": 1.76e-05,
|
| 162 |
+
"loss": 0.7107,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.9028459273797841,
|
| 167 |
+
"grad_norm": 16.375,
|
| 168 |
+
"learning_rate": 1.8400000000000003e-05,
|
| 169 |
+
"loss": 1.2491,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.9421000981354269,
|
| 174 |
+
"grad_norm": 35.5,
|
| 175 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 176 |
+
"loss": 0.6448,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.9813542688910697,
|
| 181 |
+
"grad_norm": 22.875,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 0.5682,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 1.0206084396467126,
|
| 188 |
+
"grad_norm": 32.75,
|
| 189 |
+
"learning_rate": 1.9999025240093045e-05,
|
| 190 |
+
"loss": 0.5689,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.0598626104023552,
|
| 195 |
+
"grad_norm": 28.375,
|
| 196 |
+
"learning_rate": 1.9996101150403543e-05,
|
| 197 |
+
"loss": 0.4836,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.099116781157998,
|
| 202 |
+
"grad_norm": 18.25,
|
| 203 |
+
"learning_rate": 1.9991228300988586e-05,
|
| 204 |
+
"loss": 0.519,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.138370951913641,
|
| 209 |
+
"grad_norm": 26.125,
|
| 210 |
+
"learning_rate": 1.9984407641819812e-05,
|
| 211 |
+
"loss": 0.4941,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.1776251226692835,
|
| 216 |
+
"grad_norm": 20.5,
|
| 217 |
+
"learning_rate": 1.9975640502598243e-05,
|
| 218 |
+
"loss": 0.4484,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 1.2168792934249264,
|
| 223 |
+
"grad_norm": 24.625,
|
| 224 |
+
"learning_rate": 1.9964928592495046e-05,
|
| 225 |
+
"loss": 0.5169,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 1.2561334641805693,
|
| 230 |
+
"grad_norm": 14.125,
|
| 231 |
+
"learning_rate": 1.9952273999818312e-05,
|
| 232 |
+
"loss": 0.4417,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.295387634936212,
|
| 237 |
+
"grad_norm": 26.25,
|
| 238 |
+
"learning_rate": 1.9937679191605964e-05,
|
| 239 |
+
"loss": 0.4361,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.3346418056918548,
|
| 244 |
+
"grad_norm": 23.125,
|
| 245 |
+
"learning_rate": 1.9921147013144782e-05,
|
| 246 |
+
"loss": 0.4454,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 1.3738959764474976,
|
| 251 |
+
"grad_norm": 23.625,
|
| 252 |
+
"learning_rate": 1.9902680687415704e-05,
|
| 253 |
+
"loss": 0.446,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.4131501472031402,
|
| 258 |
+
"grad_norm": 13.9375,
|
| 259 |
+
"learning_rate": 1.988228381446553e-05,
|
| 260 |
+
"loss": 0.4006,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 1.452404317958783,
|
| 265 |
+
"grad_norm": 31.5,
|
| 266 |
+
"learning_rate": 1.985996037070505e-05,
|
| 267 |
+
"loss": 0.4577,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 1.491658488714426,
|
| 272 |
+
"grad_norm": 47.5,
|
| 273 |
+
"learning_rate": 1.983571470813386e-05,
|
| 274 |
+
"loss": 0.4207,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.5309126594700686,
|
| 279 |
+
"grad_norm": 12.3125,
|
| 280 |
+
"learning_rate": 1.9809551553491918e-05,
|
| 281 |
+
"loss": 0.4251,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.5701668302257115,
|
| 286 |
+
"grad_norm": 20.75,
|
| 287 |
+
"learning_rate": 1.9781476007338058e-05,
|
| 288 |
+
"loss": 0.3968,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 1.6094210009813543,
|
| 293 |
+
"grad_norm": 15.0,
|
| 294 |
+
"learning_rate": 1.9751493543055634e-05,
|
| 295 |
+
"loss": 0.4019,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.648675171736997,
|
| 300 |
+
"grad_norm": 11.5,
|
| 301 |
+
"learning_rate": 1.9719610005785466e-05,
|
| 302 |
+
"loss": 0.3894,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 1.6879293424926398,
|
| 307 |
+
"grad_norm": 20.375,
|
| 308 |
+
"learning_rate": 1.9685831611286312e-05,
|
| 309 |
+
"loss": 0.3923,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 1.7271835132482827,
|
| 314 |
+
"grad_norm": 16.75,
|
| 315 |
+
"learning_rate": 1.9650164944723116e-05,
|
| 316 |
+
"loss": 0.374,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.7664376840039253,
|
| 321 |
+
"grad_norm": 21.375,
|
| 322 |
+
"learning_rate": 1.961261695938319e-05,
|
| 323 |
+
"loss": 0.402,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.8056918547595682,
|
| 328 |
+
"grad_norm": 14.25,
|
| 329 |
+
"learning_rate": 1.9573194975320672e-05,
|
| 330 |
+
"loss": 0.3646,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 1.844946025515211,
|
| 335 |
+
"grad_norm": 15.0625,
|
| 336 |
+
"learning_rate": 1.9531906677929472e-05,
|
| 337 |
+
"loss": 0.3451,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.8842001962708537,
|
| 342 |
+
"grad_norm": 12.75,
|
| 343 |
+
"learning_rate": 1.9488760116444966e-05,
|
| 344 |
+
"loss": 0.3373,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 1.9234543670264965,
|
| 349 |
+
"grad_norm": 12.5,
|
| 350 |
+
"learning_rate": 1.944376370237481e-05,
|
| 351 |
+
"loss": 0.3693,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 1.9627085377821394,
|
| 356 |
+
"grad_norm": 2.34375,
|
| 357 |
+
"learning_rate": 1.9396926207859085e-05,
|
| 358 |
+
"loss": 0.3595,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 2.001962708537782,
|
| 363 |
+
"grad_norm": 19.75,
|
| 364 |
+
"learning_rate": 1.9348256763960146e-05,
|
| 365 |
+
"loss": 0.3795,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 2.041216879293425,
|
| 370 |
+
"grad_norm": 9.5625,
|
| 371 |
+
"learning_rate": 1.9297764858882516e-05,
|
| 372 |
+
"loss": 0.3388,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 2.0804710500490677,
|
| 377 |
+
"grad_norm": 9.625,
|
| 378 |
+
"learning_rate": 1.9245460336123136e-05,
|
| 379 |
+
"loss": 0.3306,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.1197252208047104,
|
| 384 |
+
"grad_norm": 25.25,
|
| 385 |
+
"learning_rate": 1.9191353392552346e-05,
|
| 386 |
+
"loss": 0.3471,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 2.1589793915603535,
|
| 391 |
+
"grad_norm": 12.6875,
|
| 392 |
+
"learning_rate": 1.913545457642601e-05,
|
| 393 |
+
"loss": 0.3268,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.198233562315996,
|
| 398 |
+
"grad_norm": 18.375,
|
| 399 |
+
"learning_rate": 1.907777478532909e-05,
|
| 400 |
+
"loss": 0.3138,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 2.2374877330716387,
|
| 405 |
+
"grad_norm": 11.375,
|
| 406 |
+
"learning_rate": 1.901832526405114e-05,
|
| 407 |
+
"loss": 0.3271,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 2.276741903827282,
|
| 412 |
+
"grad_norm": 3.484375,
|
| 413 |
+
"learning_rate": 1.895711760239413e-05,
|
| 414 |
+
"loss": 0.3189,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 2.3159960745829244,
|
| 419 |
+
"grad_norm": 21.0,
|
| 420 |
+
"learning_rate": 1.889416373291298e-05,
|
| 421 |
+
"loss": 0.3413,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.355250245338567,
|
| 426 |
+
"grad_norm": 12.1875,
|
| 427 |
+
"learning_rate": 1.8829475928589272e-05,
|
| 428 |
+
"loss": 0.3222,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 2.39450441609421,
|
| 433 |
+
"grad_norm": 4.03125,
|
| 434 |
+
"learning_rate": 1.8763066800438638e-05,
|
| 435 |
+
"loss": 0.3238,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 2.433758586849853,
|
| 440 |
+
"grad_norm": 8.5625,
|
| 441 |
+
"learning_rate": 1.869494929505219e-05,
|
| 442 |
+
"loss": 0.3133,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 2.4730127576054954,
|
| 447 |
+
"grad_norm": 22.25,
|
| 448 |
+
"learning_rate": 1.8625136692072577e-05,
|
| 449 |
+
"loss": 0.3337,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 2.5122669283611385,
|
| 454 |
+
"grad_norm": 6.28125,
|
| 455 |
+
"learning_rate": 1.855364260160507e-05,
|
| 456 |
+
"loss": 0.3016,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 2.551521099116781,
|
| 461 |
+
"grad_norm": 8.0,
|
| 462 |
+
"learning_rate": 1.848048096156426e-05,
|
| 463 |
+
"loss": 0.3138,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.590775269872424,
|
| 468 |
+
"grad_norm": 13.5,
|
| 469 |
+
"learning_rate": 1.8405666034956842e-05,
|
| 470 |
+
"loss": 0.295,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.630029440628067,
|
| 475 |
+
"grad_norm": 18.125,
|
| 476 |
+
"learning_rate": 1.8329212407100996e-05,
|
| 477 |
+
"loss": 0.3002,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.6692836113837095,
|
| 482 |
+
"grad_norm": 17.375,
|
| 483 |
+
"learning_rate": 1.8251134982782952e-05,
|
| 484 |
+
"loss": 0.3026,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 2.708537782139352,
|
| 489 |
+
"grad_norm": 8.75,
|
| 490 |
+
"learning_rate": 1.8171448983351284e-05,
|
| 491 |
+
"loss": 0.2936,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 2.7477919528949952,
|
| 496 |
+
"grad_norm": 13.375,
|
| 497 |
+
"learning_rate": 1.8090169943749477e-05,
|
| 498 |
+
"loss": 0.3181,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 2.787046123650638,
|
| 503 |
+
"grad_norm": 22.0,
|
| 504 |
+
"learning_rate": 1.8007313709487334e-05,
|
| 505 |
+
"loss": 0.2988,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.8263002944062805,
|
| 510 |
+
"grad_norm": 10.0,
|
| 511 |
+
"learning_rate": 1.792289643355191e-05,
|
| 512 |
+
"loss": 0.3096,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.8655544651619236,
|
| 517 |
+
"grad_norm": 21.25,
|
| 518 |
+
"learning_rate": 1.78369345732584e-05,
|
| 519 |
+
"loss": 0.2987,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.904808635917566,
|
| 524 |
+
"grad_norm": 3.84375,
|
| 525 |
+
"learning_rate": 1.7749444887041797e-05,
|
| 526 |
+
"loss": 0.2787,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 2.944062806673209,
|
| 531 |
+
"grad_norm": 10.4375,
|
| 532 |
+
"learning_rate": 1.766044443118978e-05,
|
| 533 |
+
"loss": 0.2763,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.983316977428852,
|
| 538 |
+
"grad_norm": 12.6875,
|
| 539 |
+
"learning_rate": 1.7569950556517566e-05,
|
| 540 |
+
"loss": 0.2841,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 3.0225711481844946,
|
| 545 |
+
"grad_norm": 8.25,
|
| 546 |
+
"learning_rate": 1.747798090498532e-05,
|
| 547 |
+
"loss": 0.2799,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 3.061825318940137,
|
| 552 |
+
"grad_norm": 15.5,
|
| 553 |
+
"learning_rate": 1.7384553406258842e-05,
|
| 554 |
+
"loss": 0.2664,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 3.1010794896957803,
|
| 559 |
+
"grad_norm": 19.125,
|
| 560 |
+
"learning_rate": 1.7289686274214116e-05,
|
| 561 |
+
"loss": 0.262,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 3.140333660451423,
|
| 566 |
+
"grad_norm": 12.875,
|
| 567 |
+
"learning_rate": 1.7193398003386514e-05,
|
| 568 |
+
"loss": 0.2588,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 3.1795878312070656,
|
| 573 |
+
"grad_norm": 4.71875,
|
| 574 |
+
"learning_rate": 1.709570736536521e-05,
|
| 575 |
+
"loss": 0.264,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 3.2188420019627086,
|
| 580 |
+
"grad_norm": 5.375,
|
| 581 |
+
"learning_rate": 1.6996633405133656e-05,
|
| 582 |
+
"loss": 0.2622,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 3.2580961727183513,
|
| 587 |
+
"grad_norm": 8.9375,
|
| 588 |
+
"learning_rate": 1.68961954373567e-05,
|
| 589 |
+
"loss": 0.2576,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.297350343473994,
|
| 594 |
+
"grad_norm": 6.4375,
|
| 595 |
+
"learning_rate": 1.6794413042615168e-05,
|
| 596 |
+
"loss": 0.2602,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 3.336604514229637,
|
| 601 |
+
"grad_norm": 6.34375,
|
| 602 |
+
"learning_rate": 1.6691306063588583e-05,
|
| 603 |
+
"loss": 0.2495,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 3.3758586849852796,
|
| 608 |
+
"grad_norm": 7.4375,
|
| 609 |
+
"learning_rate": 1.6586894601186804e-05,
|
| 610 |
+
"loss": 0.2568,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 3.4151128557409223,
|
| 615 |
+
"grad_norm": 12.25,
|
| 616 |
+
"learning_rate": 1.6481199010631312e-05,
|
| 617 |
+
"loss": 0.2429,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 3.4543670264965654,
|
| 622 |
+
"grad_norm": 12.875,
|
| 623 |
+
"learning_rate": 1.63742398974869e-05,
|
| 624 |
+
"loss": 0.2647,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 3.493621197252208,
|
| 629 |
+
"grad_norm": 9.625,
|
| 630 |
+
"learning_rate": 1.6266038113644605e-05,
|
| 631 |
+
"loss": 0.2571,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.5328753680078506,
|
| 636 |
+
"grad_norm": 5.0625,
|
| 637 |
+
"learning_rate": 1.6156614753256583e-05,
|
| 638 |
+
"loss": 0.2565,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 3.5721295387634937,
|
| 643 |
+
"grad_norm": 11.4375,
|
| 644 |
+
"learning_rate": 1.6045991148623752e-05,
|
| 645 |
+
"loss": 0.2591,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 3.6113837095191363,
|
| 650 |
+
"grad_norm": 7.71875,
|
| 651 |
+
"learning_rate": 1.5934188866037017e-05,
|
| 652 |
+
"loss": 0.2364,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 3.650637880274779,
|
| 657 |
+
"grad_norm": 12.75,
|
| 658 |
+
"learning_rate": 1.5821229701572897e-05,
|
| 659 |
+
"loss": 0.2508,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 3.689892051030422,
|
| 664 |
+
"grad_norm": 12.375,
|
| 665 |
+
"learning_rate": 1.570713567684432e-05,
|
| 666 |
+
"loss": 0.2251,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 3.7291462217860647,
|
| 671 |
+
"grad_norm": 4.5,
|
| 672 |
+
"learning_rate": 1.5591929034707468e-05,
|
| 673 |
+
"loss": 0.2359,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 3.7684003925417073,
|
| 678 |
+
"grad_norm": 8.625,
|
| 679 |
+
"learning_rate": 1.5475632234925505e-05,
|
| 680 |
+
"loss": 0.2498,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 3.8076545632973504,
|
| 685 |
+
"grad_norm": 8.625,
|
| 686 |
+
"learning_rate": 1.5358267949789968e-05,
|
| 687 |
+
"loss": 0.2337,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 3.846908734052993,
|
| 692 |
+
"grad_norm": 8.1875,
|
| 693 |
+
"learning_rate": 1.5239859059700794e-05,
|
| 694 |
+
"loss": 0.254,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 3.8861629048086357,
|
| 699 |
+
"grad_norm": 7.0625,
|
| 700 |
+
"learning_rate": 1.5120428648705716e-05,
|
| 701 |
+
"loss": 0.249,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 3.9254170755642788,
|
| 706 |
+
"grad_norm": 3.640625,
|
| 707 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 708 |
+
"loss": 0.2464,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 3.9646712463199214,
|
| 713 |
+
"grad_norm": 13.5625,
|
| 714 |
+
"learning_rate": 1.4878596591387329e-05,
|
| 715 |
+
"loss": 0.2452,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.003925417075564,
|
| 720 |
+
"grad_norm": 20.0,
|
| 721 |
+
"learning_rate": 1.4756242090702756e-05,
|
| 722 |
+
"loss": 0.2338,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 4.043179587831207,
|
| 727 |
+
"grad_norm": 5.65625,
|
| 728 |
+
"learning_rate": 1.463296035119862e-05,
|
| 729 |
+
"loss": 0.2043,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 4.08243375858685,
|
| 734 |
+
"grad_norm": 7.375,
|
| 735 |
+
"learning_rate": 1.4508775406894308e-05,
|
| 736 |
+
"loss": 0.2217,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 4.121687929342492,
|
| 741 |
+
"grad_norm": 4.0,
|
| 742 |
+
"learning_rate": 1.4383711467890776e-05,
|
| 743 |
+
"loss": 0.2288,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 4.1609421000981355,
|
| 748 |
+
"grad_norm": 6.15625,
|
| 749 |
+
"learning_rate": 1.4257792915650728e-05,
|
| 750 |
+
"loss": 0.2023,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 4.200196270853779,
|
| 755 |
+
"grad_norm": 9.25,
|
| 756 |
+
"learning_rate": 1.413104429824542e-05,
|
| 757 |
+
"loss": 0.2005,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.239450441609421,
|
| 762 |
+
"grad_norm": 12.5625,
|
| 763 |
+
"learning_rate": 1.4003490325568953e-05,
|
| 764 |
+
"loss": 0.2098,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 4.278704612365064,
|
| 769 |
+
"grad_norm": 14.4375,
|
| 770 |
+
"learning_rate": 1.3875155864521031e-05,
|
| 771 |
+
"loss": 0.2059,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 4.317958783120707,
|
| 776 |
+
"grad_norm": 7.5,
|
| 777 |
+
"learning_rate": 1.3746065934159123e-05,
|
| 778 |
+
"loss": 0.1923,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 4.357212953876349,
|
| 783 |
+
"grad_norm": 4.53125,
|
| 784 |
+
"learning_rate": 1.3616245700820922e-05,
|
| 785 |
+
"loss": 0.2082,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.396467124631992,
|
| 790 |
+
"grad_norm": 1.9375,
|
| 791 |
+
"learning_rate": 1.3485720473218153e-05,
|
| 792 |
+
"loss": 0.213,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 4.435721295387635,
|
| 797 |
+
"grad_norm": 6.59375,
|
| 798 |
+
"learning_rate": 1.3354515697502552e-05,
|
| 799 |
+
"loss": 0.2213,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.4749754661432775,
|
| 804 |
+
"grad_norm": 10.1875,
|
| 805 |
+
"learning_rate": 1.3222656952305113e-05,
|
| 806 |
+
"loss": 0.1971,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 4.5142296368989205,
|
| 811 |
+
"grad_norm": 3.9375,
|
| 812 |
+
"learning_rate": 1.3090169943749475e-05,
|
| 813 |
+
"loss": 0.2035,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 4.553483807654564,
|
| 818 |
+
"grad_norm": 3.171875,
|
| 819 |
+
"learning_rate": 1.2957080500440469e-05,
|
| 820 |
+
"loss": 0.1887,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 4.592737978410206,
|
| 825 |
+
"grad_norm": 5.15625,
|
| 826 |
+
"learning_rate": 1.2823414568428767e-05,
|
| 827 |
+
"loss": 0.1914,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 4.631992149165849,
|
| 832 |
+
"grad_norm": 3.0,
|
| 833 |
+
"learning_rate": 1.2689198206152657e-05,
|
| 834 |
+
"loss": 0.2014,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 4.671246319921492,
|
| 839 |
+
"grad_norm": 3.3125,
|
| 840 |
+
"learning_rate": 1.2554457579357906e-05,
|
| 841 |
+
"loss": 0.1934,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 4.710500490677134,
|
| 846 |
+
"grad_norm": 4.46875,
|
| 847 |
+
"learning_rate": 1.2419218955996677e-05,
|
| 848 |
+
"loss": 0.1914,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 4.749754661432777,
|
| 853 |
+
"grad_norm": 5.65625,
|
| 854 |
+
"learning_rate": 1.2283508701106559e-05,
|
| 855 |
+
"loss": 0.2069,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 4.78900883218842,
|
| 860 |
+
"grad_norm": 2.078125,
|
| 861 |
+
"learning_rate": 1.2147353271670634e-05,
|
| 862 |
+
"loss": 0.2163,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 4.8282630029440625,
|
| 867 |
+
"grad_norm": 5.25,
|
| 868 |
+
"learning_rate": 1.2010779211459649e-05,
|
| 869 |
+
"loss": 0.2,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 4.867517173699706,
|
| 874 |
+
"grad_norm": 10.375,
|
| 875 |
+
"learning_rate": 1.187381314585725e-05,
|
| 876 |
+
"loss": 0.1848,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 4.906771344455349,
|
| 881 |
+
"grad_norm": 3.828125,
|
| 882 |
+
"learning_rate": 1.1736481776669307e-05,
|
| 883 |
+
"loss": 0.1913,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 4.946025515210991,
|
| 888 |
+
"grad_norm": 5.1875,
|
| 889 |
+
"learning_rate": 1.159881187691835e-05,
|
| 890 |
+
"loss": 0.1963,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 4.985279685966634,
|
| 895 |
+
"grad_norm": 7.96875,
|
| 896 |
+
"learning_rate": 1.1460830285624119e-05,
|
| 897 |
+
"loss": 0.1911,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 5.024533856722277,
|
| 902 |
+
"grad_norm": 3.703125,
|
| 903 |
+
"learning_rate": 1.1322563902571227e-05,
|
| 904 |
+
"loss": 0.1775,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 5.063788027477919,
|
| 909 |
+
"grad_norm": 5.84375,
|
| 910 |
+
"learning_rate": 1.1184039683065014e-05,
|
| 911 |
+
"loss": 0.1645,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 5.103042198233562,
|
| 916 |
+
"grad_norm": 5.40625,
|
| 917 |
+
"learning_rate": 1.1045284632676535e-05,
|
| 918 |
+
"loss": 0.1624,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 5.142296368989205,
|
| 923 |
+
"grad_norm": 8.3125,
|
| 924 |
+
"learning_rate": 1.0906325801977804e-05,
|
| 925 |
+
"loss": 0.1761,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.181550539744848,
|
| 930 |
+
"grad_norm": 2.09375,
|
| 931 |
+
"learning_rate": 1.0767190281268187e-05,
|
| 932 |
+
"loss": 0.1761,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 5.220804710500491,
|
| 937 |
+
"grad_norm": 4.71875,
|
| 938 |
+
"learning_rate": 1.0627905195293135e-05,
|
| 939 |
+
"loss": 0.1744,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 5.260058881256134,
|
| 944 |
+
"grad_norm": 8.125,
|
| 945 |
+
"learning_rate": 1.0488497697956134e-05,
|
| 946 |
+
"loss": 0.1787,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 5.299313052011776,
|
| 951 |
+
"grad_norm": 5.6875,
|
| 952 |
+
"learning_rate": 1.0348994967025012e-05,
|
| 953 |
+
"loss": 0.1651,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 5.338567222767419,
|
| 958 |
+
"grad_norm": 5.1875,
|
| 959 |
+
"learning_rate": 1.0209424198833571e-05,
|
| 960 |
+
"loss": 0.1684,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 5.377821393523062,
|
| 965 |
+
"grad_norm": 5.71875,
|
| 966 |
+
"learning_rate": 1.0069812602979617e-05,
|
| 967 |
+
"loss": 0.1699,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 5.417075564278704,
|
| 972 |
+
"grad_norm": 4.78125,
|
| 973 |
+
"learning_rate": 9.930187397020385e-06,
|
| 974 |
+
"loss": 0.1677,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 5.456329735034347,
|
| 979 |
+
"grad_norm": 3.0,
|
| 980 |
+
"learning_rate": 9.790575801166432e-06,
|
| 981 |
+
"loss": 0.1612,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 5.4955839057899905,
|
| 986 |
+
"grad_norm": 2.21875,
|
| 987 |
+
"learning_rate": 9.651005032974994e-06,
|
| 988 |
+
"loss": 0.1623,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 5.534838076545633,
|
| 993 |
+
"grad_norm": 1.515625,
|
| 994 |
+
"learning_rate": 9.511502302043867e-06,
|
| 995 |
+
"loss": 0.1674,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 5.574092247301276,
|
| 1000 |
+
"grad_norm": 2.390625,
|
| 1001 |
+
"learning_rate": 9.372094804706867e-06,
|
| 1002 |
+
"loss": 0.1629,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 5.613346418056919,
|
| 1007 |
+
"grad_norm": 5.59375,
|
| 1008 |
+
"learning_rate": 9.232809718731815e-06,
|
| 1009 |
+
"loss": 0.1625,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 5.652600588812561,
|
| 1014 |
+
"grad_norm": 1.453125,
|
| 1015 |
+
"learning_rate": 9.093674198022201e-06,
|
| 1016 |
+
"loss": 0.1667,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 5.691854759568204,
|
| 1021 |
+
"grad_norm": 2.640625,
|
| 1022 |
+
"learning_rate": 8.954715367323468e-06,
|
| 1023 |
+
"loss": 0.1702,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 5.731108930323847,
|
| 1028 |
+
"grad_norm": 1.3125,
|
| 1029 |
+
"learning_rate": 8.815960316934991e-06,
|
| 1030 |
+
"loss": 0.1646,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 5.770363101079489,
|
| 1035 |
+
"grad_norm": 3.359375,
|
| 1036 |
+
"learning_rate": 8.677436097428775e-06,
|
| 1037 |
+
"loss": 0.1635,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 5.809617271835132,
|
| 1042 |
+
"grad_norm": 5.3125,
|
| 1043 |
+
"learning_rate": 8.539169714375885e-06,
|
| 1044 |
+
"loss": 0.1533,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 5.8488714425907755,
|
| 1049 |
+
"grad_norm": 1.7890625,
|
| 1050 |
+
"learning_rate": 8.401188123081653e-06,
|
| 1051 |
+
"loss": 0.1535,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 5.888125613346418,
|
| 1056 |
+
"grad_norm": 3.453125,
|
| 1057 |
+
"learning_rate": 8.263518223330698e-06,
|
| 1058 |
+
"loss": 0.1586,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 5.927379784102061,
|
| 1063 |
+
"grad_norm": 2.28125,
|
| 1064 |
+
"learning_rate": 8.126186854142752e-06,
|
| 1065 |
+
"loss": 0.17,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 5.966633954857704,
|
| 1070 |
+
"grad_norm": 2.5,
|
| 1071 |
+
"learning_rate": 7.989220788540356e-06,
|
| 1072 |
+
"loss": 0.1553,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 6.005888125613346,
|
| 1077 |
+
"grad_norm": 1.25,
|
| 1078 |
+
"learning_rate": 7.852646728329368e-06,
|
| 1079 |
+
"loss": 0.145,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 6.045142296368989,
|
| 1084 |
+
"grad_norm": 2.765625,
|
| 1085 |
+
"learning_rate": 7.716491298893443e-06,
|
| 1086 |
+
"loss": 0.1419,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 6.084396467124632,
|
| 1091 |
+
"grad_norm": 5.5625,
|
| 1092 |
+
"learning_rate": 7.580781044003324e-06,
|
| 1093 |
+
"loss": 0.1442,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 6.123650637880274,
|
| 1098 |
+
"grad_norm": 4.71875,
|
| 1099 |
+
"learning_rate": 7.445542420642097e-06,
|
| 1100 |
+
"loss": 0.1397,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 6.1629048086359175,
|
| 1105 |
+
"grad_norm": 2.359375,
|
| 1106 |
+
"learning_rate": 7.310801793847344e-06,
|
| 1107 |
+
"loss": 0.1453,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 6.202158979391561,
|
| 1112 |
+
"grad_norm": 3.0625,
|
| 1113 |
+
"learning_rate": 7.176585431571235e-06,
|
| 1114 |
+
"loss": 0.1383,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 6.241413150147203,
|
| 1119 |
+
"grad_norm": 2.171875,
|
| 1120 |
+
"learning_rate": 7.042919499559538e-06,
|
| 1121 |
+
"loss": 0.1474,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 6.280667320902846,
|
| 1126 |
+
"grad_norm": 1.0859375,
|
| 1127 |
+
"learning_rate": 6.909830056250527e-06,
|
| 1128 |
+
"loss": 0.1485,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 6.319921491658489,
|
| 1133 |
+
"grad_norm": 2.515625,
|
| 1134 |
+
"learning_rate": 6.777343047694891e-06,
|
| 1135 |
+
"loss": 0.1457,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 6.359175662414131,
|
| 1140 |
+
"grad_norm": 5.5,
|
| 1141 |
+
"learning_rate": 6.645484302497452e-06,
|
| 1142 |
+
"loss": 0.1458,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 6.398429833169774,
|
| 1147 |
+
"grad_norm": 1.453125,
|
| 1148 |
+
"learning_rate": 6.5142795267818505e-06,
|
| 1149 |
+
"loss": 0.1425,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 6.437684003925417,
|
| 1154 |
+
"grad_norm": 3.8125,
|
| 1155 |
+
"learning_rate": 6.383754299179079e-06,
|
| 1156 |
+
"loss": 0.1405,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 6.4769381746810595,
|
| 1161 |
+
"grad_norm": 1.1875,
|
| 1162 |
+
"learning_rate": 6.25393406584088e-06,
|
| 1163 |
+
"loss": 0.1386,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 6.516192345436703,
|
| 1168 |
+
"grad_norm": 1.7109375,
|
| 1169 |
+
"learning_rate": 6.124844135478971e-06,
|
| 1170 |
+
"loss": 0.1444,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 6.555446516192346,
|
| 1175 |
+
"grad_norm": 1.28125,
|
| 1176 |
+
"learning_rate": 5.996509674431053e-06,
|
| 1177 |
+
"loss": 0.1341,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.594700686947988,
|
| 1182 |
+
"grad_norm": 0.890625,
|
| 1183 |
+
"learning_rate": 5.868955701754584e-06,
|
| 1184 |
+
"loss": 0.1479,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 6.633954857703631,
|
| 1189 |
+
"grad_norm": 0.9140625,
|
| 1190 |
+
"learning_rate": 5.742207084349274e-06,
|
| 1191 |
+
"loss": 0.1399,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 6.673209028459274,
|
| 1196 |
+
"grad_norm": 3.34375,
|
| 1197 |
+
"learning_rate": 5.616288532109225e-06,
|
| 1198 |
+
"loss": 0.1424,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 6.712463199214916,
|
| 1203 |
+
"grad_norm": 1.171875,
|
| 1204 |
+
"learning_rate": 5.491224593105695e-06,
|
| 1205 |
+
"loss": 0.1358,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 6.751717369970559,
|
| 1210 |
+
"grad_norm": 2.140625,
|
| 1211 |
+
"learning_rate": 5.367039648801386e-06,
|
| 1212 |
+
"loss": 0.1385,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 6.790971540726202,
|
| 1217 |
+
"grad_norm": 1.03125,
|
| 1218 |
+
"learning_rate": 5.243757909297247e-06,
|
| 1219 |
+
"loss": 0.1356,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.8302257114818445,
|
| 1224 |
+
"grad_norm": 1.96875,
|
| 1225 |
+
"learning_rate": 5.121403408612672e-06,
|
| 1226 |
+
"loss": 0.1349,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 6.869479882237488,
|
| 1231 |
+
"grad_norm": 1.5,
|
| 1232 |
+
"learning_rate": 5.000000000000003e-06,
|
| 1233 |
+
"loss": 0.1376,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 6.908734052993131,
|
| 1238 |
+
"grad_norm": 4.25,
|
| 1239 |
+
"learning_rate": 4.879571351294287e-06,
|
| 1240 |
+
"loss": 0.143,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 6.947988223748773,
|
| 1245 |
+
"grad_norm": 1.2109375,
|
| 1246 |
+
"learning_rate": 4.76014094029921e-06,
|
| 1247 |
+
"loss": 0.1417,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 6.987242394504416,
|
| 1252 |
+
"grad_norm": 1.265625,
|
| 1253 |
+
"learning_rate": 4.641732050210032e-06,
|
| 1254 |
+
"loss": 0.1371,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 7.026496565260059,
|
| 1259 |
+
"grad_norm": 0.9375,
|
| 1260 |
+
"learning_rate": 4.524367765074499e-06,
|
| 1261 |
+
"loss": 0.1392,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 7.065750736015701,
|
| 1266 |
+
"grad_norm": 1.9140625,
|
| 1267 |
+
"learning_rate": 4.408070965292534e-06,
|
| 1268 |
+
"loss": 0.1295,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 7.105004906771344,
|
| 1273 |
+
"grad_norm": 0.8984375,
|
| 1274 |
+
"learning_rate": 4.292864323155684e-06,
|
| 1275 |
+
"loss": 0.126,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 7.144259077526987,
|
| 1280 |
+
"grad_norm": 0.80078125,
|
| 1281 |
+
"learning_rate": 4.178770298427107e-06,
|
| 1282 |
+
"loss": 0.1215,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.18351324828263,
|
| 1287 |
+
"grad_norm": 1.1953125,
|
| 1288 |
+
"learning_rate": 4.065811133962987e-06,
|
| 1289 |
+
"loss": 0.1375,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 7.222767419038273,
|
| 1294 |
+
"grad_norm": 1.734375,
|
| 1295 |
+
"learning_rate": 3.954008851376252e-06,
|
| 1296 |
+
"loss": 0.1284,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 7.262021589793916,
|
| 1301 |
+
"grad_norm": 1.2421875,
|
| 1302 |
+
"learning_rate": 3.8433852467434175e-06,
|
| 1303 |
+
"loss": 0.1303,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 7.301275760549558,
|
| 1308 |
+
"grad_norm": 0.8984375,
|
| 1309 |
+
"learning_rate": 3.7339618863553983e-06,
|
| 1310 |
+
"loss": 0.1273,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 7.340529931305201,
|
| 1315 |
+
"grad_norm": 0.89453125,
|
| 1316 |
+
"learning_rate": 3.625760102513103e-06,
|
| 1317 |
+
"loss": 0.1303,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 7.379784102060844,
|
| 1322 |
+
"grad_norm": 2.265625,
|
| 1323 |
+
"learning_rate": 3.5188009893686916e-06,
|
| 1324 |
+
"loss": 0.1321,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.419038272816486,
|
| 1329 |
+
"grad_norm": 1.6484375,
|
| 1330 |
+
"learning_rate": 3.4131053988131947e-06,
|
| 1331 |
+
"loss": 0.1284,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 7.458292443572129,
|
| 1336 |
+
"grad_norm": 0.7109375,
|
| 1337 |
+
"learning_rate": 3.308693936411421e-06,
|
| 1338 |
+
"loss": 0.1376,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 7.4975466143277725,
|
| 1343 |
+
"grad_norm": 1.7421875,
|
| 1344 |
+
"learning_rate": 3.2055869573848374e-06,
|
| 1345 |
+
"loss": 0.1271,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 7.536800785083415,
|
| 1350 |
+
"grad_norm": 0.69921875,
|
| 1351 |
+
"learning_rate": 3.103804562643302e-06,
|
| 1352 |
+
"loss": 0.1264,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 7.576054955839058,
|
| 1357 |
+
"grad_norm": 1.234375,
|
| 1358 |
+
"learning_rate": 3.003366594866345e-06,
|
| 1359 |
+
"loss": 0.1293,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 7.615309126594701,
|
| 1364 |
+
"grad_norm": 0.98046875,
|
| 1365 |
+
"learning_rate": 2.9042926346347932e-06,
|
| 1366 |
+
"loss": 0.1272,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 7.654563297350343,
|
| 1371 |
+
"grad_norm": 0.87109375,
|
| 1372 |
+
"learning_rate": 2.8066019966134907e-06,
|
| 1373 |
+
"loss": 0.1332,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 7.693817468105986,
|
| 1378 |
+
"grad_norm": 0.83984375,
|
| 1379 |
+
"learning_rate": 2.7103137257858867e-06,
|
| 1380 |
+
"loss": 0.134,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 7.733071638861629,
|
| 1385 |
+
"grad_norm": 1.1015625,
|
| 1386 |
+
"learning_rate": 2.615446593741161e-06,
|
| 1387 |
+
"loss": 0.1314,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 7.772325809617271,
|
| 1392 |
+
"grad_norm": 1.2265625,
|
| 1393 |
+
"learning_rate": 2.522019095014683e-06,
|
| 1394 |
+
"loss": 0.1194,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 7.8115799803729145,
|
| 1399 |
+
"grad_norm": 0.78515625,
|
| 1400 |
+
"learning_rate": 2.4300494434824373e-06,
|
| 1401 |
+
"loss": 0.1306,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 7.8508341511285575,
|
| 1406 |
+
"grad_norm": 1.171875,
|
| 1407 |
+
"learning_rate": 2.339555568810221e-06,
|
| 1408 |
+
"loss": 0.1278,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 7.8900883218842,
|
| 1413 |
+
"grad_norm": 0.9375,
|
| 1414 |
+
"learning_rate": 2.2505551129582047e-06,
|
| 1415 |
+
"loss": 0.1328,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 7.929342492639843,
|
| 1420 |
+
"grad_norm": 1.0390625,
|
| 1421 |
+
"learning_rate": 2.163065426741603e-06,
|
| 1422 |
+
"loss": 0.1272,
|
| 1423 |
+
"step": 2020
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 7.968596663395486,
|
| 1427 |
+
"grad_norm": 0.70703125,
|
| 1428 |
+
"learning_rate": 2.0771035664480944e-06,
|
| 1429 |
+
"loss": 0.127,
|
| 1430 |
+
"step": 2030
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 8.007850834151128,
|
| 1434 |
+
"grad_norm": 0.640625,
|
| 1435 |
+
"learning_rate": 1.9926862905126663e-06,
|
| 1436 |
+
"loss": 0.123,
|
| 1437 |
+
"step": 2040
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 8.047105004906772,
|
| 1441 |
+
"grad_norm": 0.78125,
|
| 1442 |
+
"learning_rate": 1.9098300562505266e-06,
|
| 1443 |
+
"loss": 0.1222,
|
| 1444 |
+
"step": 2050
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 8.086359175662414,
|
| 1448 |
+
"grad_norm": 0.765625,
|
| 1449 |
+
"learning_rate": 1.8285510166487154e-06,
|
| 1450 |
+
"loss": 0.1277,
|
| 1451 |
+
"step": 2060
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 8.125613346418056,
|
| 1455 |
+
"grad_norm": 0.85546875,
|
| 1456 |
+
"learning_rate": 1.7488650172170496e-06,
|
| 1457 |
+
"loss": 0.1255,
|
| 1458 |
+
"step": 2070
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 8.1648675171737,
|
| 1462 |
+
"grad_norm": 0.74609375,
|
| 1463 |
+
"learning_rate": 1.6707875928990059e-06,
|
| 1464 |
+
"loss": 0.1249,
|
| 1465 |
+
"step": 2080
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 8.204121687929343,
|
| 1469 |
+
"grad_norm": 0.6875,
|
| 1470 |
+
"learning_rate": 1.5943339650431578e-06,
|
| 1471 |
+
"loss": 0.1222,
|
| 1472 |
+
"step": 2090
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 8.243375858684985,
|
| 1476 |
+
"grad_norm": 0.8828125,
|
| 1477 |
+
"learning_rate": 1.5195190384357405e-06,
|
| 1478 |
+
"loss": 0.1223,
|
| 1479 |
+
"step": 2100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 8.282630029440629,
|
| 1483 |
+
"grad_norm": 0.9375,
|
| 1484 |
+
"learning_rate": 1.446357398394934e-06,
|
| 1485 |
+
"loss": 0.126,
|
| 1486 |
+
"step": 2110
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 8.321884200196271,
|
| 1490 |
+
"grad_norm": 0.875,
|
| 1491 |
+
"learning_rate": 1.3748633079274254e-06,
|
| 1492 |
+
"loss": 0.1312,
|
| 1493 |
+
"step": 2120
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 8.361138370951913,
|
| 1497 |
+
"grad_norm": 0.796875,
|
| 1498 |
+
"learning_rate": 1.30505070494781e-06,
|
| 1499 |
+
"loss": 0.125,
|
| 1500 |
+
"step": 2130
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 8.400392541707557,
|
| 1504 |
+
"grad_norm": 0.7578125,
|
| 1505 |
+
"learning_rate": 1.2369331995613664e-06,
|
| 1506 |
+
"loss": 0.1311,
|
| 1507 |
+
"step": 2140
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 8.4396467124632,
|
| 1511 |
+
"grad_norm": 1.3046875,
|
| 1512 |
+
"learning_rate": 1.1705240714107301e-06,
|
| 1513 |
+
"loss": 0.1281,
|
| 1514 |
+
"step": 2150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 8.478900883218841,
|
| 1518 |
+
"grad_norm": 0.5546875,
|
| 1519 |
+
"learning_rate": 1.1058362670870248e-06,
|
| 1520 |
+
"loss": 0.1209,
|
| 1521 |
+
"step": 2160
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 8.518155053974485,
|
| 1525 |
+
"grad_norm": 1.296875,
|
| 1526 |
+
"learning_rate": 1.042882397605871e-06,
|
| 1527 |
+
"loss": 0.1274,
|
| 1528 |
+
"step": 2170
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 8.557409224730128,
|
| 1532 |
+
"grad_norm": 1.265625,
|
| 1533 |
+
"learning_rate": 9.816747359488632e-07,
|
| 1534 |
+
"loss": 0.1237,
|
| 1535 |
+
"step": 2180
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 8.59666339548577,
|
| 1539 |
+
"grad_norm": 0.82421875,
|
| 1540 |
+
"learning_rate": 9.222252146709143e-07,
|
| 1541 |
+
"loss": 0.1254,
|
| 1542 |
+
"step": 2190
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 8.635917566241414,
|
| 1546 |
+
"grad_norm": 0.84375,
|
| 1547 |
+
"learning_rate": 8.645454235739903e-07,
|
| 1548 |
+
"loss": 0.1301,
|
| 1549 |
+
"step": 2200
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 8.675171736997056,
|
| 1553 |
+
"grad_norm": 0.76171875,
|
| 1554 |
+
"learning_rate": 8.086466074476562e-07,
|
| 1555 |
+
"loss": 0.1248,
|
| 1556 |
+
"step": 2210
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 8.714425907752698,
|
| 1560 |
+
"grad_norm": 0.65234375,
|
| 1561 |
+
"learning_rate": 7.545396638768698e-07,
|
| 1562 |
+
"loss": 0.1205,
|
| 1563 |
+
"step": 2220
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 8.753680078508342,
|
| 1567 |
+
"grad_norm": 0.6640625,
|
| 1568 |
+
"learning_rate": 7.022351411174866e-07,
|
| 1569 |
+
"loss": 0.1231,
|
| 1570 |
+
"step": 2230
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 8.792934249263984,
|
| 1574 |
+
"grad_norm": 1.109375,
|
| 1575 |
+
"learning_rate": 6.517432360398556e-07,
|
| 1576 |
+
"loss": 0.1206,
|
| 1577 |
+
"step": 2240
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 8.832188420019627,
|
| 1581 |
+
"grad_norm": 0.9375,
|
| 1582 |
+
"learning_rate": 6.030737921409169e-07,
|
| 1583 |
+
"loss": 0.1261,
|
| 1584 |
+
"step": 2250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 8.87144259077527,
|
| 1588 |
+
"grad_norm": 0.78125,
|
| 1589 |
+
"learning_rate": 5.562362976251901e-07,
|
| 1590 |
+
"loss": 0.1297,
|
| 1591 |
+
"step": 2260
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 8.910696761530913,
|
| 1595 |
+
"grad_norm": 0.734375,
|
| 1596 |
+
"learning_rate": 5.112398835550348e-07,
|
| 1597 |
+
"loss": 0.129,
|
| 1598 |
+
"step": 2270
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 8.949950932286555,
|
| 1602 |
+
"grad_norm": 0.61328125,
|
| 1603 |
+
"learning_rate": 4.6809332207053083e-07,
|
| 1604 |
+
"loss": 0.1233,
|
| 1605 |
+
"step": 2280
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 8.989205103042199,
|
| 1609 |
+
"grad_norm": 1.1015625,
|
| 1610 |
+
"learning_rate": 4.268050246793276e-07,
|
| 1611 |
+
"loss": 0.1225,
|
| 1612 |
+
"step": 2290
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 9.028459273797841,
|
| 1616 |
+
"grad_norm": 0.85546875,
|
| 1617 |
+
"learning_rate": 3.8738304061681107e-07,
|
| 1618 |
+
"loss": 0.126,
|
| 1619 |
+
"step": 2300
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 9.067713444553483,
|
| 1623 |
+
"grad_norm": 0.74609375,
|
| 1624 |
+
"learning_rate": 3.498350552768859e-07,
|
| 1625 |
+
"loss": 0.1229,
|
| 1626 |
+
"step": 2310
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 9.106967615309127,
|
| 1630 |
+
"grad_norm": 0.91015625,
|
| 1631 |
+
"learning_rate": 3.1416838871368925e-07,
|
| 1632 |
+
"loss": 0.1247,
|
| 1633 |
+
"step": 2320
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 9.14622178606477,
|
| 1637 |
+
"grad_norm": 0.765625,
|
| 1638 |
+
"learning_rate": 2.8038999421453827e-07,
|
| 1639 |
+
"loss": 0.1241,
|
| 1640 |
+
"step": 2330
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 9.185475956820412,
|
| 1644 |
+
"grad_norm": 1.015625,
|
| 1645 |
+
"learning_rate": 2.4850645694436736e-07,
|
| 1646 |
+
"loss": 0.1279,
|
| 1647 |
+
"step": 2340
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 9.224730127576056,
|
| 1651 |
+
"grad_norm": 0.734375,
|
| 1652 |
+
"learning_rate": 2.1852399266194312e-07,
|
| 1653 |
+
"loss": 0.1256,
|
| 1654 |
+
"step": 2350
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 9.263984298331698,
|
| 1658 |
+
"grad_norm": 0.66796875,
|
| 1659 |
+
"learning_rate": 1.9044844650808468e-07,
|
| 1660 |
+
"loss": 0.1283,
|
| 1661 |
+
"step": 2360
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 9.30323846908734,
|
| 1665 |
+
"grad_norm": 0.63671875,
|
| 1666 |
+
"learning_rate": 1.6428529186614195e-07,
|
| 1667 |
+
"loss": 0.1246,
|
| 1668 |
+
"step": 2370
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 9.342492639842984,
|
| 1672 |
+
"grad_norm": 0.79296875,
|
| 1673 |
+
"learning_rate": 1.400396292949513e-07,
|
| 1674 |
+
"loss": 0.1263,
|
| 1675 |
+
"step": 2380
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 9.381746810598626,
|
| 1679 |
+
"grad_norm": 1.5,
|
| 1680 |
+
"learning_rate": 1.1771618553447217e-07,
|
| 1681 |
+
"loss": 0.1323,
|
| 1682 |
+
"step": 2390
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 9.421000981354268,
|
| 1686 |
+
"grad_norm": 0.90625,
|
| 1687 |
+
"learning_rate": 9.731931258429638e-08,
|
| 1688 |
+
"loss": 0.1307,
|
| 1689 |
+
"step": 2400
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 9.460255152109912,
|
| 1693 |
+
"grad_norm": 0.91796875,
|
| 1694 |
+
"learning_rate": 7.885298685522235e-08,
|
| 1695 |
+
"loss": 0.1202,
|
| 1696 |
+
"step": 2410
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 9.499509322865554,
|
| 1700 |
+
"grad_norm": 0.67578125,
|
| 1701 |
+
"learning_rate": 6.232080839403631e-08,
|
| 1702 |
+
"loss": 0.1277,
|
| 1703 |
+
"step": 2420
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 9.538763493621197,
|
| 1707 |
+
"grad_norm": 0.77734375,
|
| 1708 |
+
"learning_rate": 4.772600018168816e-08,
|
| 1709 |
+
"loss": 0.1303,
|
| 1710 |
+
"step": 2430
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 9.57801766437684,
|
| 1714 |
+
"grad_norm": 0.9140625,
|
| 1715 |
+
"learning_rate": 3.50714075049563e-08,
|
| 1716 |
+
"loss": 0.1238,
|
| 1717 |
+
"step": 2440
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 9.617271835132483,
|
| 1721 |
+
"grad_norm": 0.75,
|
| 1722 |
+
"learning_rate": 2.4359497401758026e-08,
|
| 1723 |
+
"loss": 0.1286,
|
| 1724 |
+
"step": 2450
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 9.656526005888125,
|
| 1728 |
+
"grad_norm": 0.62890625,
|
| 1729 |
+
"learning_rate": 1.5592358180189782e-08,
|
| 1730 |
+
"loss": 0.1165,
|
| 1731 |
+
"step": 2460
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 9.695780176643769,
|
| 1735 |
+
"grad_norm": 1.109375,
|
| 1736 |
+
"learning_rate": 8.771699011416169e-09,
|
| 1737 |
+
"loss": 0.1244,
|
| 1738 |
+
"step": 2470
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 9.735034347399411,
|
| 1742 |
+
"grad_norm": 0.94140625,
|
| 1743 |
+
"learning_rate": 3.898849596456477e-09,
|
| 1744 |
+
"loss": 0.1258,
|
| 1745 |
+
"step": 2480
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 9.774288518155053,
|
| 1749 |
+
"grad_norm": 0.70703125,
|
| 1750 |
+
"learning_rate": 9.74759906957612e-10,
|
| 1751 |
+
"loss": 0.1216,
|
| 1752 |
+
"step": 2490
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 9.813542688910697,
|
| 1756 |
+
"grad_norm": 0.7265625,
|
| 1757 |
+
"learning_rate": 0.0,
|
| 1758 |
+
"loss": 0.1174,
|
| 1759 |
+
"step": 2500
|
| 1760 |
+
}
|
| 1761 |
+
],
|
| 1762 |
+
"logging_steps": 10,
|
| 1763 |
+
"max_steps": 2500,
|
| 1764 |
+
"num_input_tokens_seen": 0,
|
| 1765 |
+
"num_train_epochs": 10,
|
| 1766 |
+
"save_steps": 500,
|
| 1767 |
+
"stateful_callbacks": {
|
| 1768 |
+
"TrainerControl": {
|
| 1769 |
+
"args": {
|
| 1770 |
+
"should_epoch_stop": false,
|
| 1771 |
+
"should_evaluate": false,
|
| 1772 |
+
"should_log": false,
|
| 1773 |
+
"should_save": true,
|
| 1774 |
+
"should_training_stop": true
|
| 1775 |
+
},
|
| 1776 |
+
"attributes": {}
|
| 1777 |
+
}
|
| 1778 |
+
},
|
| 1779 |
+
"total_flos": 9.5664155394048e+17,
|
| 1780 |
+
"train_batch_size": 4,
|
| 1781 |
+
"trial_name": null,
|
| 1782 |
+
"trial_params": null
|
| 1783 |
+
}
|
checkpoint-2500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a04ba1fea5935bbaf15347f768d6702bcc3e5630f2593cba25a017368a06442
|
| 3 |
+
size 5432
|